nokogiri 1.13.6 → 1.16.7

data/ext/nokogiri/xml_document_fragment.c

@@ -19,7 +19,7 @@ new (int argc, VALUE *argv, VALUE klass)
 
   rb_scan_args(argc, argv, "1*", &document, &rest);
 
-  Data_Get_Struct(document, xmlDoc, xml_doc);
+  xml_doc = noko_xml_document_unwrap(document);
 
   node = xmlNewDocFragment(xml_doc->doc);
 
@@ -32,7 +32,7 @@ new (int argc, VALUE *argv, VALUE klass)
 }
 
 void
-noko_init_xml_document_fragment()
+noko_init_xml_document_fragment(void)
 {
   assert(cNokogiriXmlNode);
   /*

data/ext/nokogiri/xml_dtd.c

@@ -44,7 +44,7 @@ entities(VALUE self)
   xmlDtdPtr dtd;
   VALUE hash;
 
-  Data_Get_Struct(self, xmlDtd, dtd);
+  Noko_Node_Get_Struct(self, xmlDtd, dtd);
 
   if (!dtd->entities) { return Qnil; }
 
@@ -67,7 +67,7 @@ notations(VALUE self)
   xmlDtdPtr dtd;
   VALUE hash;
 
-  Data_Get_Struct(self, xmlDtd, dtd);
+  Noko_Node_Get_Struct(self, xmlDtd, dtd);
 
   if (!dtd->notations) { return Qnil; }
 
@@ -90,7 +90,7 @@ attributes(VALUE self)
   xmlDtdPtr dtd;
   VALUE hash;
 
-  Data_Get_Struct(self, xmlDtd, dtd);
+  Noko_Node_Get_Struct(self, xmlDtd, dtd);
 
   hash = rb_hash_new();
 
@@ -113,7 +113,7 @@ elements(VALUE self)
   xmlDtdPtr dtd;
   VALUE hash;
 
-  Data_Get_Struct(self, xmlDtd, dtd);
+  Noko_Node_Get_Struct(self, xmlDtd, dtd);
 
   if (!dtd->elements) { return Qnil; }
 
@@ -138,8 +138,8 @@ validate(VALUE self, VALUE document)
   xmlValidCtxtPtr ctxt;
   VALUE error_list;
 
-  Data_Get_Struct(self, xmlDtd, dtd);
-  Data_Get_Struct(document, xmlDoc, doc);
+  Noko_Node_Get_Struct(self, xmlDtd, dtd);
+  doc = noko_xml_document_unwrap(document);
   error_list = rb_ary_new();
 
   ctxt = xmlNewValidCtxt();
@@ -165,7 +165,7 @@ static VALUE
 system_id(VALUE self)
 {
   xmlDtdPtr dtd;
-  Data_Get_Struct(self, xmlDtd, dtd);
+  Noko_Node_Get_Struct(self, xmlDtd, dtd);
 
   if (!dtd->SystemID) { return Qnil; }
 
@@ -182,7 +182,7 @@ static VALUE
 external_id(VALUE self)
 {
   xmlDtdPtr dtd;
-  Data_Get_Struct(self, xmlDtd, dtd);
+  Noko_Node_Get_Struct(self, xmlDtd, dtd);
 
   if (!dtd->ExternalID) { return Qnil; }
 
@@ -190,7 +190,7 @@ external_id(VALUE self)
 }
 
 void
-noko_init_xml_dtd()
+noko_init_xml_dtd(void)
 {
   assert(cNokogiriXmlNode);
   /*

data/ext/nokogiri/xml_element_content.c

@@ -2,17 +2,22 @@
 
 VALUE cNokogiriXmlElementContent;
 
+static const rb_data_type_t element_content_data_type = {
+  .wrap_struct_name = "Nokogiri::XML::ElementContent",
+  .flags = RUBY_TYPED_FREE_IMMEDIATELY | RUBY_TYPED_WB_PROTECTED,
+};
+
 /*
  * call-seq:
- *  name
+ *   name → String
  *
- * Get the require element +name+
+ * [Returns] The content element's +name+
  */
 static VALUE
 get_name(VALUE self)
 {
   xmlElementContentPtr elem;
-  Data_Get_Struct(self, xmlElementContent, elem);
+  TypedData_Get_Struct(self, xmlElementContent, &element_content_data_type, elem);
 
   if (!elem->name) { return Qnil; }
   return NOKOGIRI_STR_NEW2(elem->name);
@@ -20,47 +25,40 @@ get_name(VALUE self)
 
 /*
  * call-seq:
- *  type
+ *   type → Integer
  *
- * Get the element content +type+.  Possible values are PCDATA, ELEMENT, SEQ,
- * or OR.
+ * [Returns] The content element's +type+. Possible values are +PCDATA+, +ELEMENT+, +SEQ+, or +OR+.
  */
 static VALUE
 get_type(VALUE self)
 {
   xmlElementContentPtr elem;
-  Data_Get_Struct(self, xmlElementContent, elem);
+  TypedData_Get_Struct(self, xmlElementContent, &element_content_data_type, elem);
 
-  return INT2NUM((long)elem->type);
+  return INT2NUM(elem->type);
 }
 
 /*
- * call-seq:
- *  c1
- *
  * Get the first child.
  */
 static VALUE
 get_c1(VALUE self)
 {
   xmlElementContentPtr elem;
-  Data_Get_Struct(self, xmlElementContent, elem);
+  TypedData_Get_Struct(self, xmlElementContent, &element_content_data_type, elem);
 
   if (!elem->c1) { return Qnil; }
   return noko_xml_element_content_wrap(rb_iv_get(self, "@document"), elem->c1);
 }
 
 /*
- * call-seq:
- *  c2
- *
- * Get the first child.
+ * Get the second child.
  */
 static VALUE
 get_c2(VALUE self)
 {
   xmlElementContentPtr elem;
-  Data_Get_Struct(self, xmlElementContent, elem);
+  TypedData_Get_Struct(self, xmlElementContent, &element_content_data_type, elem);
 
   if (!elem->c2) { return Qnil; }
   return noko_xml_element_content_wrap(rb_iv_get(self, "@document"), elem->c2);
@@ -68,51 +66,56 @@ get_c2(VALUE self)
 
 /*
  * call-seq:
- *  occur
+ *   occur → Integer
  *
- * Get the element content +occur+ flag.  Possible values are ONCE, OPT, MULT
- * or PLUS.
+ * [Returns] The content element's +occur+ flag. Possible values are +ONCE+, +OPT+, +MULT+ or +PLUS+.
  */
 static VALUE
 get_occur(VALUE self)
 {
   xmlElementContentPtr elem;
-  Data_Get_Struct(self, xmlElementContent, elem);
+  TypedData_Get_Struct(self, xmlElementContent, &element_content_data_type, elem);
 
-  return INT2NUM((long)elem->ocur);
+  return INT2NUM(elem->ocur);
 }
 
 /*
  * call-seq:
- *  prefix
+ *   prefix → String
  *
- * Get the element content namespace +prefix+.
+ * [Returns] The content element's namespace +prefix+.
  */
 static VALUE
 get_prefix(VALUE self)
 {
   xmlElementContentPtr elem;
-  Data_Get_Struct(self, xmlElementContent, elem);
+  TypedData_Get_Struct(self, xmlElementContent, &element_content_data_type, elem);
 
   if (!elem->prefix) { return Qnil; }
 
   return NOKOGIRI_STR_NEW2(elem->prefix);
 }
 
+/*
+ *  create a Nokogiri::XML::ElementContent object around an +element+.
+ */
 VALUE
-noko_xml_element_content_wrap(VALUE doc, xmlElementContentPtr element)
+noko_xml_element_content_wrap(VALUE rb_document, xmlElementContentPtr c_element_content)
 {
-  VALUE elem = Data_Wrap_Struct(cNokogiriXmlElementContent, 0, 0, element);
+  VALUE elem = TypedData_Wrap_Struct(
+                 cNokogiriXmlElementContent,
+                 &element_content_data_type,
+                 c_element_content
+               );
 
-  /* Setting the document is necessary so that this does not get GC'd until */
-  /* the document is GC'd */
-  rb_iv_set(elem, "@document", doc);
+  /* keep a handle on the document for GC marking */
+  rb_iv_set(elem, "@document", rb_document);
 
   return elem;
 }
 
 void
-noko_init_xml_element_content()
+noko_init_xml_element_content(void)
 {
   cNokogiriXmlElementContent = rb_define_class_under(mNokogiriXml, "ElementContent", rb_cObject);
 

data/ext/nokogiri/xml_element_decl.c

@@ -6,7 +6,7 @@ static ID id_document;
 
 /*
  * call-seq:
- *  element_type
+ *   element_type → Integer
  *
  * The element_type
  */
@@ -14,21 +14,21 @@ static VALUE
 element_type(VALUE self)
 {
   xmlElementPtr node;
-  Data_Get_Struct(self, xmlElement, node);
-  return INT2NUM((long)node->etype);
+  Noko_Node_Get_Struct(self, xmlElement, node);
+  return INT2NUM(node->etype);
 }
 
 /*
  * call-seq:
- *  content
+ *   content → Nokogiri::XML::ElementContent
  *
- * The allowed content for this ElementDecl
+ * [Returns] The root of this element declaration's content tree.
  */
 static VALUE
 content(VALUE self)
 {
   xmlElementPtr node;
-  Data_Get_Struct(self, xmlElement, node);
+  Noko_Node_Get_Struct(self, xmlElement, node);
 
   if (!node->content) { return Qnil; }
 
@@ -40,15 +40,15 @@ content(VALUE self)
 
 /*
  * call-seq:
- *  prefix
+ *   prefix → String
  *
- * The namespace prefix for this ElementDecl
+ * [Returns] The namespace +prefix+ for this element declaration.
  */
 static VALUE
 prefix(VALUE self)
 {
   xmlElementPtr node;
-  Data_Get_Struct(self, xmlElement, node);
+  Noko_Node_Get_Struct(self, xmlElement, node);
 
   if (!node->prefix) { return Qnil; }
 
@@ -56,7 +56,7 @@ prefix(VALUE self)
 }
 
 void
-noko_init_xml_element_decl()
+noko_init_xml_element_decl(void)
 {
   assert(cNokogiriXmlNode);
   cNokogiriXmlElementDecl = rb_define_class_under(mNokogiriXml, "ElementDecl", cNokogiriXmlNode);

data/ext/nokogiri/xml_encoding_handler.c

@@ -2,14 +2,22 @@
 
 VALUE cNokogiriEncodingHandler;
 
-
 static void
-_xml_encoding_handler_dealloc(xmlCharEncodingHandlerPtr c_handler)
+xml_encoding_handler_dealloc(void *data)
 {
   /* make sure iconv handlers are cleaned up and freed */
+  xmlCharEncodingHandlerPtr c_handler = data;
   xmlCharEncCloseFunc(c_handler);
 }
 
+static const rb_data_type_t xml_encoding_handler_type = {
+  .wrap_struct_name = "Nokogiri::EncodingHandler",
+  .function = {
+    .dfree = xml_encoding_handler_dealloc,
+  },
+  .flags = RUBY_TYPED_FREE_IMMEDIATELY | RUBY_TYPED_WB_PROTECTED,
+};
+
 
 /*
  * call-seq: Nokogiri::EncodingHandler.[](name)
@@ -23,7 +31,7 @@ rb_xml_encoding_handler_s_get(VALUE klass, VALUE key)
 
   handler = xmlFindCharEncodingHandler(StringValueCStr(key));
   if (handler) {
-    return Data_Wrap_Struct(klass, NULL, _xml_encoding_handler_dealloc, handler);
+    return TypedData_Wrap_Struct(klass, &xml_encoding_handler_type, handler);
   }
 
   return Qnil;
@@ -45,9 +53,9 @@ rb_xml_encoding_handler_s_delete(VALUE klass, VALUE name)
 
 
 /*
- * call-seq: Nokogiri::EncodingHandler.alias(from, to)
+ * call-seq: Nokogiri::EncodingHandler.alias(real_name, alias_name)
  *
- * Alias encoding handler with name +from+ to name +to+
+ * Alias encoding handler with name +real_name+ to name +alias_name+
  */
 static VALUE
 rb_xml_encoding_handler_s_alias(VALUE klass, VALUE from, VALUE to)
@@ -82,14 +90,14 @@ rb_xml_encoding_handler_name(VALUE self)
 {
   xmlCharEncodingHandlerPtr handler;
 
-  Data_Get_Struct(self, xmlCharEncodingHandler, handler);
+  TypedData_Get_Struct(self, xmlCharEncodingHandler, &xml_encoding_handler_type, handler);
 
   return NOKOGIRI_STR_NEW2(handler->name);
 }
 
 
 void
-noko_init_xml_encoding_handler()
+noko_init_xml_encoding_handler(void)
 {
   cNokogiriEncodingHandler = rb_define_class_under(mNokogiri, "EncodingHandler", rb_cObject);
 

data/ext/nokogiri/xml_entity_decl.c

@@ -12,7 +12,7 @@ static VALUE
 original_content(VALUE self)
 {
   xmlEntityPtr node;
-  Data_Get_Struct(self, xmlEntity, node);
+  Noko_Node_Get_Struct(self, xmlEntity, node);
 
   if (!node->orig) { return Qnil; }
 
@@ -29,7 +29,7 @@ static VALUE
 get_content(VALUE self)
 {
   xmlEntityPtr node;
-  Data_Get_Struct(self, xmlEntity, node);
+  Noko_Node_Get_Struct(self, xmlEntity, node);
 
   if (!node->content) { return Qnil; }
 
@@ -46,7 +46,7 @@ static VALUE
 entity_type(VALUE self)
 {
   xmlEntityPtr node;
-  Data_Get_Struct(self, xmlEntity, node);
+  Noko_Node_Get_Struct(self, xmlEntity, node);
 
   return INT2NUM((int)node->etype);
 }
@@ -61,7 +61,7 @@ static VALUE
 external_id(VALUE self)
 {
   xmlEntityPtr node;
-  Data_Get_Struct(self, xmlEntity, node);
+  Noko_Node_Get_Struct(self, xmlEntity, node);
 
   if (!node->ExternalID) { return Qnil; }
 
@@ -78,7 +78,7 @@ static VALUE
 system_id(VALUE self)
 {
   xmlEntityPtr node;
-  Data_Get_Struct(self, xmlEntity, node);
+  Noko_Node_Get_Struct(self, xmlEntity, node);
 
   if (!node->SystemID) { return Qnil; }
 
@@ -86,7 +86,7 @@ system_id(VALUE self)
 }
 
 void
-noko_init_xml_entity_decl()
+noko_init_xml_entity_decl(void)
 {
   assert(cNokogiriXmlNode);
   cNokogiriXmlEntityDecl = rb_define_class_under(mNokogiriXml, "EntityDecl", cNokogiriXmlNode);

data/ext/nokogiri/xml_entity_reference.c

@@ -20,7 +20,7 @@ new (int argc, VALUE *argv, VALUE klass)
 
   rb_scan_args(argc, argv, "2*", &document, &name, &rest);
 
-  Data_Get_Struct(document, xmlDoc, xml_doc);
+  xml_doc = noko_xml_document_unwrap(document);
 
   node = xmlNewReference(
            xml_doc,
@@ -38,7 +38,7 @@ new (int argc, VALUE *argv, VALUE klass)
 }
 
 void
-noko_init_xml_entity_reference()
+noko_init_xml_entity_reference(void)
 {
   assert(cNokogiriXmlNode);
   /*

data/ext/nokogiri/xml_namespace.c

@@ -25,13 +25,14 @@
 VALUE cNokogiriXmlNamespace ;
 
 static void
-dealloc_namespace(xmlNsPtr ns)
+_xml_namespace_dealloc(void *ptr)
 {
   /*
    * this deallocator is only used for namespace nodes that are part of an xpath
    * node set. see noko_xml_namespace_wrap().
    */
-  NOKOGIRI_DEBUG_START(ns) ;
+  xmlNsPtr ns = ptr;
+
   if (ns->href) {
     xmlFree(DISCARD_CONST_QUAL_XMLCHAR(ns->href));
   }
@@ -39,39 +40,95 @@ dealloc_namespace(xmlNsPtr ns)
     xmlFree(DISCARD_CONST_QUAL_XMLCHAR(ns->prefix));
   }
   xmlFree(ns);
-  NOKOGIRI_DEBUG_END(ns) ;
 }
 
+static void
+_xml_namespace_update_references(void *ptr)
+{
+  xmlNsPtr ns = ptr;
+  if (ns->_private) {
+    ns->_private = (void *)rb_gc_location((VALUE)ns->_private);
+  }
+}
+
+static const rb_data_type_t nokogiri_xml_namespace_type_with_dealloc = {
+  .wrap_struct_name = "Nokogiri::XML::Namespace#with_dealloc",
+  .function = {
+    .dfree = _xml_namespace_dealloc,
+    .dcompact = _xml_namespace_update_references,
+  },
+  .flags = RUBY_TYPED_FREE_IMMEDIATELY | RUBY_TYPED_WB_PROTECTED,
+};
+
+static const rb_data_type_t nokogiri_xml_namespace_type_without_dealloc = {
+  .wrap_struct_name = "Nokogiri::XML::Namespace#without_dealloc",
+  .function = {
+    .dcompact = _xml_namespace_update_references,
+  },
+  .flags = RUBY_TYPED_FREE_IMMEDIATELY | RUBY_TYPED_WB_PROTECTED,
+};
 
 /*
- * call-seq:
- *  prefix
+ *  :call-seq:
+ *    prefix() → String or nil
+ *
+ *  Return the prefix for this Namespace, or +nil+ if there is no prefix (e.g., default namespace).
  *
- * Get the prefix for this namespace.  Returns +nil+ if there is no prefix.
+ *  *Example*
+ *
+ *    doc = Nokogiri::XML.parse(<<~XML)
+ *      <?xml version="1.0"?>
+ *      <root xmlns="http://nokogiri.org/ns/default" xmlns:noko="http://nokogiri.org/ns/noko">
+ *        <child1 foo="abc" noko:bar="def"/>
+ *        <noko:child2 foo="qwe" noko:bar="rty"/>
+ *      </root>
+ *    XML
+ *
+ *    doc.root.elements.first.namespace.prefix
+ *    # => nil
+ *
+ *    doc.root.elements.last.namespace.prefix
+ *    # => "noko"
  */
 static VALUE
 prefix(VALUE self)
 {
   xmlNsPtr ns;
 
-  Data_Get_Struct(self, xmlNs, ns);
+  Noko_Namespace_Get_Struct(self, xmlNs, ns);
   if (!ns->prefix) { return Qnil; }
 
   return NOKOGIRI_STR_NEW2(ns->prefix);
 }
 
 /*
- * call-seq:
- *  href
+ *  :call-seq:
+ *    href() → String
+ *
+ *  Returns the URI reference for this Namespace.
+ *
+ *  *Example*
+ *
+ *    doc = Nokogiri::XML.parse(<<~XML)
+ *      <?xml version="1.0"?>
+ *      <root xmlns="http://nokogiri.org/ns/default" xmlns:noko="http://nokogiri.org/ns/noko">
+ *        <child1 foo="abc" noko:bar="def"/>
+ *        <noko:child2 foo="qwe" noko:bar="rty"/>
+ *      </root>
+ *    XML
+ *
+ *    doc.root.elements.first.namespace.href
+ *    # => "http://nokogiri.org/ns/default"
  *
- * Get the href for this namespace
+ *    doc.root.elements.last.namespace.href
+ *    # => "http://nokogiri.org/ns/noko"
  */
 static VALUE
 href(VALUE self)
 {
   xmlNsPtr ns;
 
-  Data_Get_Struct(self, xmlNs, ns);
+  Noko_Namespace_Get_Struct(self, xmlNs, ns);
   if (!ns->href) { return Qnil; }
 
   return NOKOGIRI_STR_NEW2(ns->href);
@@ -87,14 +144,18 @@ noko_xml_namespace_wrap(xmlNsPtr c_namespace, xmlDocPtr c_document)
   }
 
   if (c_document) {
-    rb_namespace = Data_Wrap_Struct(cNokogiriXmlNamespace, 0, 0, c_namespace);
+    rb_namespace = TypedData_Wrap_Struct(cNokogiriXmlNamespace,
+                                         &nokogiri_xml_namespace_type_without_dealloc,
+                                         c_namespace);
 
     if (DOC_RUBY_OBJECT_TEST(c_document)) {
       rb_iv_set(rb_namespace, "@document", DOC_RUBY_OBJECT(c_document));
       rb_ary_push(DOC_NODE_CACHE(c_document), rb_namespace);
     }
   } else {
-    rb_namespace = Data_Wrap_Struct(cNokogiriXmlNamespace, 0, dealloc_namespace, c_namespace);
+    rb_namespace = TypedData_Wrap_Struct(cNokogiriXmlNamespace,
+                                         &nokogiri_xml_namespace_type_with_dealloc,
+                                         c_namespace);
   }
 
   c_namespace->_private = (void *)rb_namespace;
@@ -109,7 +170,7 @@ noko_xml_namespace_wrap_xpath_copy(xmlNsPtr c_namespace)
 }
 
 void
-noko_init_xml_namespace()
+noko_init_xml_namespace(void)
 {
   cNokogiriXmlNamespace = rb_define_class_under(mNokogiriXml, "Namespace", rb_cObject);